← Selected work

Selected work / Ethereum / research

Counterfact

An interface for protocol counterfactuals, model assumptions and reproducible artifact records.

The experiment index — Counterfact
Selected moments / September 2026A compact project selection.

The idea

Put the assumptions beside the result.

This selection keeps the experiment index, the fee-market study and the issuance study. The stills document the public interface as captured in September 2026.

Live and validated labels belong to the original captured interface. Historical replays and model results were not independently reproduced during this migration.

Three selected moments

A closer look.

Move through the selected stills.

The experiment index — Counterfact

01 / Selected moment

The experiment index

The experiment index from the captured Counterfact interface.

Open full image ↗
The fee-market study — Counterfact

02 / Selected moment

The fee-market study

The fee-market study from the captured Counterfact interface.

Open full image ↗
The issuance study — Counterfact

03 / Selected moment

The issuance study

The issuance study from the captured Counterfact interface.

Open full image ↗

Make something of your own

Build your own.

Take the idea into Codex and adapt it to your own subject, sources and audience.

Reconstructed recipe · Not the original prompts
An end-to-end recreation has not been tested.

Download all prompts ↓
01BriefDefine the experience

Build a new interpretation of Counterfact: An interface for protocol counterfactuals, model assumptions and reproducible artifact records. Read the repository instructions and preserve existing work first. Reconstruct the idea from this selection; do not promise an exact recreation. Start with one complete useful flow.

02IngredientsPrepare sources and assets

Retain versioned historical inputs, specification references, recurrence/formula implementations, baseline parameters and artifact provenance. Keep observed history separate from changed assumptions.

03InteractionBuild the core flow

Implement one baseline replay before parameter controls. Show units, model limits and differences from observed history. Add URL-encoded configurations and Markdown export, then compare baseline outputs with independently supplied source records.

04Verify & preserveLeave a working handoff

Check the real production route, keyboard operation, mobile layout, reduced motion, asset loading and failure states. Record what passed and what remains untested. Leave source, lockfiles, setup instructions, data provenance and unique selected assets in a compact package. Exclude secrets, dependencies and caches. Label reconstructed prompts honestly; a local package is not a NAS backup.

← All selected work